JP5027932B2 - Active pixel sensor device used for star lacquer device - Google Patents

Description

  The present invention relates to an active pixel sensor device (APS) for use in a start lacquer device, as described in the preamble of claim 1.

It is well known that a start lacquer image is damaged by a so-called single event upset (SEU) that appears as a point or streak that destroys the integrity of the image as a cosmic ray abnormality.
And single event upsets are caused by solar flares and other cosmic radiation, and are particularly troublesome, especially in activities approaching the sun.

On the other hand, in recent years, an active pixel sensor (APS) type image device chip is mounted on a start tracker device that recognizes a pattern by tracking a specific constellation or star with a camera and observes its own position. ing.
The present invention proposes that each pixel has a photodetector and an active amplifier and that further functionality is integrated into an APS sensor including an integrated circuit that is an array of each pixel sensor. is there.

An example of such an image sensor designed specifically for a starter lacquer device is the so-called low-cost, low-developed, developed by Cybless Semiconductor, in cooperation with the applicant, European Space Agency (ESA). There is a mass (LCMS) sensor.
This type of APS device provides an on-chip logic circuit that allows the on-chip logic circuit to perform both destructive readout and non-destructive readout of the pixels of the imaging device chip.

Further, the center of the pixel is formed by the accumulation amount of the photodiode generated by the irradiated photons.
Destructive readout also emits a pixel photodiode, while non-destructive readout consists in measuring the transistor's transmittance, which depends on the amount of transistor gate voltage stored in the photodiode.

As a result, the transistor acts as an amplifier (amplifier) for the pixel signal.
Currently, non-destructive readout is used to perform correlated double sampling to minimize fixed pattern noise (FPN) of the image signal.

Furthermore, Patent Document 1 (US Pat. No. 7,145,188 (B2)) discloses providing an on-chip logic circuit on a CMOS type image device chip.
That is, it has been proposed to perform local adaptation of pixel gain when one or more pixels in a region are saturated.

US Patent US 7,145,188 (B2) (full text)

Accordingly, an object of the present invention is to provide an active pixel sensor device for performing on-the-fly processing of a single event upset.
In particular, in order to reduce the complexity and complexity of image processing in the start lacquer apparatus, it is possible to perform on-the-fly rejection processing of these single event upsets.

That is, the present invention is first composed of an active pixel sensor device for a start lacquer device including a chip for an image device.
The imaging device chip includes an array of photodiodes that operate as optical pixels.
Further, the active pixel sensor device includes a logic circuit that performs the following functions.

That is, such a logic circuit can process the output of the optical pixel and output a modified signal.
Furthermore, the logic circuit is configured to read out a signal of one pixel nondestructively in accordance with the amount of light irradiated during a predetermined integration time (sometimes referred to simply as “predetermined time”). ing.
A logic circuit is provided that resets the optical pixel when the predetermined integration time ends.
In particular, the pixel is reset and then sampled, and the pixel signal can be read out twice or multiple times without additional resetting during that time.

Therefore, the main object of the present invention is to include a logic circuit that can detect whether or not a discontinuity has occurred in the pixel signal during the integration time, and modify the pixel signal according to this detection result. It is to be.
Such an idea is based on the fact that light from a permanent light source such as a star results in a continuous variation of the pixel signal, for example a linear increase in the amount stored in the photodiode.

Single event upset (SEU), on the other hand, is a phenomenon in which a pixel signal jumps suddenly during a given integration time and is considered as a discontinuity.
Thus, the detection of discontinuities is to take into account the identification of signals affected by SEU and the identification of different effects on the signal, in particular the effects of continuous light sources due to the effects of SEU.
And in order to remove the influence of SEU, it should be subtracted from the signal in the discontinuous phenomenon part.

  Also, as another embodiment of the invention, regarding the signal modification, such an active pixel sensor device can be used as an illumination sensor when the continuous part is removed instead of the discontinuous part. is there.

  Further, when the logic circuit is an on-chip logic circuit provided in the image device chip, a particularly small active pixel device suitable for a satellite can be provided.

Further, the logic circuit is configured to periodically perform nondestructive reading during a predetermined integration time .
This can allow quasi-continuous sampling of the signal in a discrete set of positions.
Also, the discontinuity phenomenon, characterized by detecting by comparing the difference between the result of the next non-destructive readout with a predetermined threshold.
That is, when such a difference exceeds the threshold value, it can be determined that the SEU has occurred between the last two non-destructive readout steps.

In a preferred embodiment, the active pixel gain is preferably set so that the brightest star in the field of view of the device generates a pixel signal just below the pixel saturation value.
Here, the threshold value must basically correspond to a value obtained by dividing the saturation value by the number of non-destructive reads performed during one integration period.

A particularly simple embodiment of the invention is when the logic circuit is configured to determine the output value by an increment that is repeatedly added to a base value (here, eg, 0): The effect of discontinuity can be eliminated.
Thus, if the previously determined difference does not reach the threshold, the difference value between the next non-destructive read result in the same integration period can be used as an increment.
On the other hand, if the previously determined difference exceeds a predetermined threshold, an extrapolated difference value can be used as an increment.

Also, if the logic circuit comprises storage means for storing at least one difference value, the logic circuit is based on the stored difference value, for example by just replicating the stored difference value, or previously The extrapolation difference value can be determined by executing a certain averaging process using the difference value determined in step (b).
And if the extrapolated difference value is determined by averaging a certain number of stored difference values, the actual stored number of previously determined difference values is sufficient to perform the averaging. If the predetermined minimum number of difference values that are considered to be not reached, the logic circuit can output a default value.

The function of the logic circuit can be supplemented by a windowed readout function of the image object.
Therein, only the pixels in a particular window covering the estimated position of the star are read out to reduce the signal processing complexity.
In addition, the logic circuit can perform correlated double sampling to avoid or reduce fixed pattern noise.
Furthermore, the background intensity level can be subtracted from the pixel signal by a logic circuit.

In each of the above-described embodiments, the aspect that the active signal processing functions can be switched on / off may be a necessary component or an optional component.
In order to allow such a selection, the active pixel sensor device can be provided with means for activating or deactivating at least one of these signal processing functions of the logic circuit.
In other words, stopping unused functions leads to a reduction in energy consumption.

Accordingly, further important features and advantages of the present invention will be described based on the following description of embodiments of the present invention, together with the accompanying drawings.
However, the present invention is not limited to the following description, and the present invention is effective even for combinations of claims that can be easily understood by those skilled in the art.

FIG. 1 shows an active pixel sensor device including a chip for an image device in a chip package suitable for a start lacquer device according to the present invention. FIG. 2 is a general schematic diagram of the chip structure of the active pixel sensor imager chip of FIG. FIG. 3 is a system level diagram of the active pixel sensor imager chip according to FIGS. FIG. 4 is a transistor level diagram of the three transistor pixels of the imaging device chip according to FIGS. FIG. 5 is a transistor level diagram functionally equivalent to the three transistor pixels of FIG. FIG. 6 is an image having a plurality of SEUs taken by an imaging device chip according to the prior art. FIG. 7 is a graph showing the change over time of the pixel signal of the imaging device chip including the SEU signal effect. FIG. 8 is a graph showing the flow of digital domain processing for correcting the pixel signal according to the result of SEU detection.

FIG. 1 shows an active pixel sensor device for a start lacquer device including a CMOS type image device chip 10.
Further, the image device chip 10 includes an array 13 of photodiodes as optical pixels.
The image device chip 10 is included in a ceramic JLCC 84 package, and has a size of 15.5 mm in the x direction and 16.2 mm in the y direction.

FIG. 2 is an overall view of the chip structure of the image device chip 10 according to FIG.
In addition to the photodiode array functioning as an optical pixel and arranged in a pixel array 13 of CMOS type 512 × 512 three transistor pixels, the imaging device chip 10 includes a logic circuit 14 having the following configuration: I have.
That is, a logic circuit 14 that includes a sample and hold column amplifier 16, a row select reset driver 18, an x-addressing shift register 20, a y-addressing register 22, and a 12-bit analog-to-digital converter 24 (ADC). It has.

Furthermore, the logic circuit 14 includes the following configuration.
That is, the read sequencer 26, the correlated double sampling and processing unit 28 provided with the memory device 30, the window sequencer 32, and the SRAM memory device 34 for storing the window sequencer 32 program.
Finally, the imaging device chip 10 includes a parallel data interface 36 and a higher level data interface 38 and is provided for communication, eg, in a “space wire” data format.

The unit of the image device chip 10 described above is disposed on the common substrate 40.
Here, the window sequencer 32 manages the window treat-out of the image device chip 10 controlled by an external IP (Internet protocol or the like) that can indicate the estimated position of the star. Means.
In addition, the logic circuit 14 uses the row select reset driver 18, the sample and hold column amplifier 16, and the shift registers 20 and 21 to read out pixels in the window type area around these estimated positions. .

The signal read from the optical pixel in the pixel array 13 is amplified using the amplifier 42, converted by the analog-digital converter 24, and read by the read sequencer 26.
The signal received from the pixel depends on the amount of light irradiated during a predetermined integration time (T).
The integration time (T) is defined as the time between the next reset process executed by the reset driver 18.

  Also, the correlated double sampling and processing unit 28 of the logic circuit 14 is configured to perform non-destructive readout of the pixel signal during the integration time (T), where the pixel signal Are read twice during the reading process without being reset.

In accordance with the present invention, the logic circuit 14 is further configured to detect whether a discontinuity has occurred in the pixel signal during the integration time (T).
According to the embodiment of the chip structure shown in FIG. 2, this configuration is recognized by a suitable program of correlated double sampling and processing unit 28.
Further, the processing unit 28 of the logic circuit 14 is configured to modify the pixel signal in response to the result of this detection, as will be described in detail below.

FIG. 3 is a system level diagram of the image device chip 10 according to the present invention.
The box on the left side of FIG. 3 shows functions performed in the image device chip 10, that is, a sequence function 42, a windowing function 44, a processing function 46, and an interface function 48.

Here, the image device chip 10 can be roughly subdivided into a sensor unit 50, an analog-digital converter unit 52, and a logic circuit unit 54 (particularly the processing unit 28) that constitutes the main part of the logic circuit 14. .
The image device chip 10 communicates with the processor 56 executing the application program 58 via the data interfaces 36 and 38.

The image device chip 10 can read out a pixel signal having an accuracy of 10 bits, and the gain and offset of the chip analog-to-digital converter 24 can be dynamically reset.
The logic circuit 14 can then read out all pixel arrays at a screen rewrite rate of 5 Hertz with an integration time T of 200 milliseconds per pixel.
Thus, in the windowed readout mode, the sensor can read out 20 windows of 20 × 20 pixels at a rate of 10 Hertz with an integration time (T) of 100 milliseconds per pixel.

  The imager chip 10 also generates all necessary timing signals and pulses to operate and control the pixel array 13 via simple input from a user including an off-chip oscillator.

The sensor also supports two modes of operation, in particular full frame readout and the windowed readout described above.
Thus, in full frame readout, the exposure and exposure start time can be specified by the user, and the imager chip 10 can read the signal of all pixels from the pixel array 13 using a roll shutter. it can.

  In the windowed readout mode, the exposure start time is specified by the user up to a maximum of 20 as non-overlapping windows selected from, for example, 10 × 10 pixels, 15 × 15 pixels, and 20 × 20 pixels. Can do.

Also, the functions of the video chip 10 and the logic circuit 14 can be easily stopped by operating any unnecessary functions via bootstrap pins or user commands. Modular design.
That is, stopping any function causes the power consumption of the image device chip 10 to be reduced.

Further, the image device chip 10 is configured such that when the radiation dose is 0, the dark current is less than 2,500 electrons per pixel per second at 25 ° C.
Thus, when the radiation dose is full, the dark current is less than 5,000 electrons per pixel per second at 25 ° C.

Also, the crosstalk between adjacent pixels is less than 5% and remains unchanged throughout life.
Furthermore, fixed pattern noise (FPN) and sensor reset noise (KTC) are less than 75 electrons per pixel and pixel response non-uniformity is less than 1.5% of all pixel arrays. 0.25% or less for all 5 × 5 pixel subarrays.
The image device chip 10 has a sensitivity defined as a product of a fill factor and a quantum efficiency of 40% or more.
Note that the spectral range of the image device chip 10 covers wavelengths from 0.4 to 0.9 microns.

  4 and 5 show in detail the pixels of the pixel array 13 of the imager chip 10.

FIG. 4 is a pixel level diagram of the pixel.
Here, each pixel includes three transistors 60, 62, and 64, and a photodiode 66.
The first transistor 60 acts as a switch 69 between the power supply and the photodiode 66.

Further, the photodiode 66 is equivalent to a capacitor 68 (see FIG. 5) having a light controlled current source.
The second transistor 62 can be regarded as a source follower amplifier 70 (see FIG. 5) that buffers the voltage at the cathode of the photodiode 66 / capacitor 68 for connection to the outside.
The third transistor 64 is a switch 72 that connects the output of the amplifier 70 to an output signal bus 74 corresponding to the column bus of the pixel array 13.

Here, in every column of pixels, two capacitors (not shown) are located for resetting the selected pixel and sampling the signal level.
Both sample capacitors are used in non-destructive readout.

Both capacitors then contain the same signal after sampling.
Also, the read sequence is the same as the destructive read, but only the signal coming from the signal sample capacitor 68 is used.
Note that the reset signal coming from the pixel is ignored.

One column amplifier 16 is provided for writing a hold signal on the sample and a hold capacitor for the output channel in all columns of the pixel.
And in destructive readout, the reset and signal levels can be subtracted at the output amplifier 16 to cancel out most of the fixed pattern noise.
Furthermore, in non-destructive readout, only the signal level is used by the output amplifier 42.

When a star lacquer device including an active pixel sensor device according to the present invention is used in an environment with a high radiation dose (eg in a satellite passing near the sun / where it is susceptible to solar flare) The energetic particles are absorbed by the photodiode, causing discontinuities in the pixel signal.
If enough energy is absorbed by the photodiode 66, the pixel saturates and a white spot occurs at the particle impact location in the pixel array 13.
These white spots disappear in the next image, but the number of points (white spots) increases with integration time.
This phenomenon is known as “single event setup (SEU)” and the star lacquer image is often damaged by SEU due to solar surface flare.

FIG. 6 shows an image of a typical start lacquer image including a plurality of SEU points and the streak 75.
Here, according to the prior art, the SEU streaks are removed by software in the image post-processing step.
However, post-processing cannot be completely removed, and is very complex and time consuming.

Then, FIG. 7 shows an example of the pixel signal during the integration time (T).
At an intermediate time (Ts), the radiation particles are absorbed by the photodiode 66, resulting in a quasi-continuous sudden jump in the pixel signal, generating an avalanche of charge carriers.

Here, it should be noted that the expressions “discontinuous” and “discontinuous” are not interpreted narrowly in a mathematical sense, but rather in terms of the typical time resolution of a digital circuit.
Thus, if the signal changes by more than 10-20% of the signal amplitude within a fraction of less than 1-2% of the total integration time T, the sudden increase or decrease of the signal is “discontinuously”. Classify as "is."

FIG. 7 shows the entire signal 72 read from the pixel, together with the influence of the dark current 74 and the influence of the target signal 76 generated from the star to the observation location.
Furthermore, FIG. 7 shows the effect of a SEU signal 78 having the form of a step function (integrated σ function) with discontinuities at Ts.

  One of the main objectives of the present invention is to perform non-destructive readout to detect discontinuities and to subtract or cancel SEU signal effects 78 by appropriate modification of the signal processing signal. It is to do.

And according to a preferred embodiment of the present invention, the logic circuit 13 is configured to perform non-destructive read repeatedly during the integration time T and from time T1 to T7.
The processing unit 28 then determines the difference between the results of the next non-destructive readout.
For example, the difference D1 in reading between T1 and T2 and D5 due to the difference in reading between T5 and T6 are determined.

Each difference D1, D5 is then compared twice with a corresponding predetermined threshold, eg, the maximum amplitude of the pixel signal divided by the number of non-destructive readouts during the integration time (T).
Furthermore, the logic circuit 13 uses the temporal response of the pixel signal 72 before SEU at Ts to extrapolate the raw signal.
If the pixel saturates after SEU, the readout value after SEU is ignored, while if the pixel is unsaturated, the readout can be corrected by subtracting the discontinuity.

In a preferred embodiment of the present invention, the logic circuit 13 is configured to determine the output signal in increments that repeatedly add to the base value. There, the base value may be zero.
If no discontinuity is detected in the previous step, for example, if there is a difference D1 that does not reach the threshold, as in the step between T1 and T2 in FIG. 7, the difference D1 can be used as an increment.
However, if the difference exceeds a predetermined threshold, for example, in the case of a step between T5 and T6 in FIG. 7, the difference D5 is detected and the logic circuit uses the extrapolated difference value as an increment.

In addition, in order to perform extrapolation, the logic circuit 13 includes a storage device 30 that stores the last number difference value.
The processing unit 28 then uses the stored difference value to determine an extrapolated difference value based on the stored difference value.

In the simplest case, only the last difference value can be stored in the storage device 30 as an extrapolation difference value.
Basically, this corresponds to approximating the discontinuous height of FIG. 7 as the difference between the difference value D5 and the difference value D4 between the non-destructive readings at T5 and T4. .

  In a more detailed embodiment of the present invention, the logic circuit may determine the extrapolated difference value by averaging two or more difference values stored in the storage device 30 and determined before detecting the jump. it can.

However, such an averaging procedure can only be performed if a predetermined minimum number of unaltered difference values is stored in the storage device 30, at least by SEU.
As a result, if the previously stored and determined difference value has not reached a predetermined minimum number that cannot be averaged, logic circuit 13 and processing unit 28 are configured to output a default value. Can.

  Non-destructive readout from T1 to T7 is performed by determining the gate voltage of the photodiode using the transistor 62 of FIG. It then acts as the source follower amplifier 70 of FIG.

  Also, the above method of extrapolating the signal uses a discretized gradient of the pixel signal that is determined in the form of a previous difference value in Ts.

  Furthermore, by applying the adjustable bias value of the analog-to-digital converter 42, the background intensity level can be subtracted from the pixel signal (eg, dark current effect 74 in FIG. 7).

Next, FIG. 8 shows an example of the digital domain processing flow performed in the processing unit 28.
That is, the processing unit 28 receives the pixel signal level from the analog-to-digital converter ADC 24 and reads the pixel signal level from the previous non-destructive readout at the same integration time T from the memory device 30 to determine the difference value. And subtract the previous signal level from the actual pixel signal level.
Further, the difference value is compared with a threshold value read from the memory device 30.

If the difference exceeds the threshold, the difference value is discarded, and the difference value stored in the previous step is read from the memory device 30 and added to the output value.
On the other hand, if the difference value does not exceed the threshold, the logic circuit concludes that the SEU did not occur in the previous step and the difference value is added to the base value and stored in the memory device 30 for the subsequent step.
If the pixel signal level received from the analog-to-digital converter 24 indicates that the pixel is saturated, the process immediately jumps to the step of adding the previous difference value.

Claims (14)

An active pixel sensor device used in a star lacquer device comprising an image device chip (10),
The imaging device chip (10) includes an array of photodiodes (13) that function as optical pixels, and a logic circuit (14);
The logic circuit (14)
Reading out the pixel signal according to the amount of light irradiated during the predetermined integration time (T) and resetting the optical pixel at the end of the predetermined integration time (T);
Processing the pixel signal and outputting a modified pixel signal;
Periodically performing a non-destructive readout of the pixel signal during said integration time (T);
Furthermore, the logic circuit (14)
Detecting whether a discontinuity has occurred in the pixel signal during the integration time (T);
Modify the pixel signal according to the detected result , where:
The logic circuit (14)
Determining the difference values [D1, D5] of successive non-destructive readout results;
An active pixel sensor device characterized in that the discontinuous phenomenon is detected by comparing a predetermined threshold value with a difference value [D1, D5] of the determined readout result ;   2. The active pixel sensor device according to claim 1, wherein the logic circuit (14) is an on-chip logic circuit provided in the imaging device chip (10). The active pixel sensor device according to claim 1 or 2 , wherein the logic circuit (14) comprises:
-The output signal value is determined by increment that is repeatedly added to the base value,
If the difference value between successive non-destructive read results is less than or equal to the predetermined threshold, use the difference value determined in advance for the difference value between successive non-destructive read results as the increment;
Using an extrapolated difference value as an increment that increases if the difference between successive non-destructive readout results exceeds a predetermined threshold value;
An active pixel sensor device. 4. The active pixel sensor device according to claim 3 , wherein the logic circuit (14) includes storage means (30) for storing at least one difference value (D4), the stored difference value ( An active pixel sensor device, wherein the extrapolation difference value is determined based on D4). The active pixel sensor device according to claim 3 or 4 , wherein the logic circuit (14) averages at least two difference values determined before detection of the jump, so that the extrapolated difference value. An active pixel sensor device, wherein the active pixel sensor device is configured to determine. 6. The active pixel sensor device according to any one of claims 3 to 5 , wherein a number of stored previously determined difference values do not reach a predetermined minimum number. (14) is configured to output a default value of a modified signal, an active pixel sensor device. The active pixel sensor device according to any one of claims 1 to 6 , wherein the photodiode is of a CMOS type. 8. The active pixel sensor device according to claim 7 , wherein the logic circuit (14) is configured to perform the non-destructive readout by determining a gate voltage of the photodiode. Active pixel sensor device. An active pixel sensor device according to any one of claims 1-8, wherein the logic circuit (14), using a gradient of a signal to be determined before jumping to perform the extrapolation, An active pixel sensor device for modifying a signal. An active pixel sensor device according to any one of claims 1 to 9, characterized in that it includes means (32) for managing a windowed readout of the imaging device chip (10) Active pixel sensor device. An active pixel sensor device according to any one of claims 1 to 10, wherein the logic circuit (14), in response to the reading of the array (13) of the photodiode, so as to perform a correlated double sampling An active pixel sensor device comprising: 12. The active pixel sensor device according to any one of claims 1 to 11 , wherein the logic circuit (14) is configured to calculate a background intensity level from a pixel signal and subtract it. An active pixel sensor device. An active pixel sensor device according to any one of claims 1 to 12 for processing at least one signal, means for the or activate the logic circuit (14), or stop (38 An active pixel sensor device. An on-chip data processing method for an active pixel sensor device for use in a star lacquer device including an imaging device chip (10) comprising:
Said imaging device chip (10) comprises an array (13) of photodiodes functioning as optical pixels;
The on-chip data processing method includes:
Reading pixel signals according to the amount of light irradiated during a predetermined integration time (T), and resetting the optical pixels upon completion of the predetermined integration time (T);
Processing the pixel signal and outputting a modified pixel signal;
Performing a non-destructive readout of the pixel signal periodically for a predetermined integration time (T);
Including
Determining the difference values [D1, D5] of successive non-destructive readout results;
By detecting a discontinuous phenomenon by comparing a predetermined threshold value and a difference value [D1, D5] of the determined readout result,
-Detecting whether said discontinuity has occurred in a pixel signal during said integration time (T);
-An on-chip data processing method further comprising the step of modifying the pixel signal in response to the detected result.

Images